David Lyon - Rule Utilitarianism

RULE-UTILITARIANISM
University Press Scholarship Online
Oxford Scholarship Online
Forms and Limits of Utilitarianism

David Lyons
Print publication date: 1965

Print ISBN-13: 9780198241973
Published to Oxford Scholarship Online: October 2011
DOI: 10.1093/acprof:oso/9780198241973.001.0001
RULE-UTILITARIANISM
David Lyons
DOI:10.1093/acprof:oso/9780198241973.003.0004
Abstract and Keywords

This chapter is about moral rules. Most proponents of utilitarian generalization
have viewed such a principle as the ground of moral rules or else have been
concerned to account for the strength of certain purported obligations that
cannot be accounted for by Act-Utilitarianism. These obligations may be
expressed in general judgements about the rightness or wrongness of acts,
judgements that can be justified by appeal to utilitarian generalization and
which can plausibly be construed as moral rules. The discussion shows how such
judgements can be generated. Connected with this topic is the relation of
utilitarian generalization to rule-utilitarianism. The discussion considers two
main types of rule-utilitarianism: primitive and ‘ideal’.
Keywords: utility, primitive rule-utilitarianism, moral rules, utilitarian generalization, Act-

Utilitarianism
Page 1 of 38
PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All
Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a
monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: Universidad
de Chile; date: 15 May 2019
RULE-UTILITARIANISM
WE shall now talk of moral rules. In the first place, most proponents of
utilitarian generalization have viewed such a principle as the ground of moral
rules or else have been concerned to account.for the strength of certain
purported obligations which (it is argued) cannot be accounted for by Act-
Utilitarianism. These obligations may be expressed in general judgements about
the rightness or wrongness of acts—judgements which can be justified by appeal
to utilitarian generalization and which can plausibly be construed as moral rules.
Part of my objective in this chapter is to show how such judgements can be
generated.
Connected with this topic is the relation of utilitarian generalization to rule-

utilitarianism. I shall discuss two main types of rule-utilitarianism: one, which I
shall call primitive rule-utilitarianism, is simply utilitarian generalization applied
in a certain way; another, typified by ‘Ideal Rule-Utilitarianism’, will be
contrasted with primitive rule-utilitarianism, and thereby with Act-Utilitarianism.
In order to discuss the various types of rule-utilitarianism, it is necessary to

distinguish different ways in which rules can be ‘grounded in’ utility. I shall
distinguish, for example, between theoretical (or theory-dependent) rules and
merely cautionary rules (rules of thumb). We may say, for the present, that
theoretical rules are rigorously justified by a given principle; they are normally
incorporated essentially into a rule-utilitarianism properly so-called; they are
rules on the basis of which the rightness or wrongness of particular acts is
determined and they often cannot be dispensed with in the application of such a
theory. Cautionary rules, on the other hand, are practical aids that are
recommended or adopted for the purpose of most efficiently and correctly
applying and acting upon a given moral principle when the principle could
theoretically be applied directly to determine the rightness or wrongness of acts.
As we shall see, (p.120) for example, the theoretical rules of Ideal Rule-
Utilitarianism cannot be dispensed with in applying that theory, for on such a
theory the rightness or wrongness of acts is not dependent upon some utilitarian
characteristic of acts themselves (such as their simple or generalized utilities),
but is rather dependent upon some utilitarian characteristic of the set of rules
under which the acts fall (its acceptance-utility).
Page 2 of 38
RULE-UTILITARIANISM
In the case of Act-Utilitarianism, by contrast, rules can theoretically be

dispensed with. For on such a theory the rightness or wrongness of an act
depends on some utilitarian characteristic of the act itself—its relative simple
utility—and no reference to rules is essential. But there are none the less at least
two ways in which rules can be associated with Act-Utilitarianism and these
must be separated for the sake of clarity. Firstly, we can have theoretical Act-
Utilitarian rules—rules that can be rigorously justified by reference to the Act-
Utilitarian principle and that indicate exactly the implications of the principle for
particular acts. A theoretical Act-Utilitarian rule against lying would have a form
like this: ‘Lying is wrong, except when its simple utility in a particular case is
greater than the simple utility of veracity, all things considered.’ This obviously
is a trivial kind of rule since in each case the rightness or wrongness of the given
act depends upon its own relative simple utility. Secondly, there are the
cautionary rules which are more commonly associated with Act-Utilitarianism—
rules that are taken as summaries of what generally or usually has been the
case, rules that indicate that acts of certain kinds are normally right or wrong,
that in the past they have usually had (relatively) good or bad effects, and that
they therefore may be expected to be right or wrong in the future. Such rules
may be employed as practical aids in deliberation and criticism, but acts are not
definitively right or wrong according to such rules so far as Act-Utilitarianism is
concerned. Because cautionary rules are general and presumably based upon
experience, they may be called ‘generalizations’. But such a label should not
suggest that these rules are based upon generalized utilities or anything of that
kind. Moreover, such rules cannot be rigorously generated from the Act-
Utilitarian principle.
Our main concern will be theoretical rules: first, those that can be generated
from (or justified by reference to) utilitarian (p.121) generalization; and
second, those incorporated into the more prominent forms of rule-utilitarianism.
Rules of the first kind are based, not on the simple utilities of acts, but rather
upon the effects of a number of acts which might be performed—upon their
generalized utilities. These are primitive rules. Whole sets of such rules justified
by a given form of utilitarian generalization yield a primitive rule-utilitarianism.
By comparing these rules with the rules of other types of rule-utilitarianism we
can determine the characteristic differences between the latter, on the one
hand, and simple and general utilitarianism on the other.
Page 3 of 38
RULE-UTILITARIANISM
The fact that a type of rule-utilitarianism can be developed from utilitarian

generalization allows us to say that nothing is essentially lost or gained—nothing
is changed—by appealing to rules as such instead of applying the test of utility
directly to acts. The forms of primitive rule-utilitarianism imply exactly what is
implied by the forms of simple utilitarianism. But we shall also find that other
types of rule-utilitarianism are not extensionally equivalent to primitive rule-
utilitarianism and therefore are not extensionally equivalent to simple
utilitarianism. A substantive alternative to Act-Utilitarianism on utilitarian
grounds is therefore possible. But the differences between these types of theory
are not obviously favourable to rule-utilitarianism. This fact will lead me to
inquire how a utilitarian can reasonably and consistently embrace a kind of rule-
utilitarianism that varies significantly from simple utilitarianism. I shall attempt
to reconstruct a mode of argument that seems to be used by utilitarians against
Act-Utilitarianism and in favour of alternative theories—a mode of argument
both the premisses and conclusions of which do not seem to be tenable.
A. Primitive Rule-Utilitarianism
It would seem that we are immediately blocked in our efforts to transform
utilitarian generalization into a kind of rule-utilitarianism. For these principles
apparently imply nothing whatever about moral rules; rules are not mentioned,
no reference is made to them in the formulations we have given the principles.
Nevertheless, from our analysis of these principles it is easy to see how a certain
mode of application of them would yield judgements which may plausibly be
called moral rules.
(p.122) One type of rule which we often think of when we speak of rules, moral
or otherwise, has the form ‘Do A’ (or ‘Don't do BY). Thus we can think of rules
such as ‘Tell the truth’, ‘Don't break promises’, and so on. Such rules could of
course be qualified in a number of ways; exceptions could explicitly be indicated,
or the grounds upon which exceptions might be determined could be mentioned;
or most simply we could add ‘other things being equal’ or an equivalent. Rules of
this rather loosely defined form may be called prescriptive (or injunctive). This is
a prominent kind of rule; but it is not the only kind.
Page 4 of 38
RULE-UTILITARIANISM
When we speak of rules we also sometimes think of what I shall call ascriptive
rules, such as ‘Lying is wrong.’ These have the basic form ′A is FY (or ′A is not F″),
which is subject to various sorts of qualifications. This is the kind of rule with
which we shall work; it is the kind most suited to the explication of a primitive
rule-utilitarianism from utilitarian generalization, largely because I have
formulated the latter principles in terms of ‘wrong’. In concentrating upon this
form of rule, I am not suggesting that this is the only or the most interesting or
the most important form of rule or of moral rule in particular; I am not
suggesting, for example, that the ascriptive form is a paradigm form for moral
rules, nor that it is essential in an adequate analysis of moral reasoning. Nor am
I suggesting that other forms of moral rules cannot be employed in the
construction of a rule-utilitarian theory, or in the development of one from
utilitarian generalization. The kinds of rules which can be employed depend
largely upon our formulation of the basic principles.
We may say in general that there is a presumption raised against any act which
satisfies a general utilitarian description (not necessarily complete) with which
is associated an undesirable tendency. It is important now that we allow
ourselves to entertain descriptions which are relevantly mcomplete, for then the
way is open to rebut such a presumption—and this is an important part of the
present programme. Suppose that we are interested, for example, in certain
given kinds of acts such as lying, the breaking of promises, black-marketeering,
walking across lawns, and so on. We can use such descriptions as starting-points
in the construction of rules. (Such a set may very likely prove to be inexhaustive,
so we shall have to add to this collection of (p.123) Starting-descriptions in
constructing a set of rules.) I shall limit my attention to acts which may be
presumed wrong.
We begin with such given, usually very general, descriptions and see if they are
relevant and whether presumptions can be established against such acts. The
presumptions can then be expressed in the main or ascriptive clauses of moral
rules, and these rules will be qualified so as to incorporate the equivalent of
rebuttals.
Take the case of lying. If the tendency of lying is undesirable then we can
presume it wrong. But lying is not necessarily—indeed not always—wrong. We
cannot consistently generate simple rules of the form ‘Lying is wrong’—without
qualification. For ‘lying’ is a description that is true of many acts, among them
acts the complete relevant descriptions of which have tendencies that are not
undesirable (or not worse than those of the alternatives, and so on—depending
upon the principle being applied). Thus, given our strong sense of ‘wrong’, we
must qualify the ascriptive rule against lying.
Page 5 of 38
RULE-UTILITARIANISM
Some qualifying clause is essential whenever such a judgement-rule is justified

by appeal to a form of utilitarian generalization because only a presumption has
been established. For some acts to which the description ‘lying’ applies, the
presumption against them will not be conclusive; for some acts this description
is significantly incomplete.
What kind of qualification? Once we speak of presumptions which can be

rebutted, the most obvious candidate is a ceteris paribus condition which is
added to and thereby qualifies the main, ascriptive, clause of the rule. This
would yield ‘If other things are equal, then lying is wrong.’ But two things must
be noted about rules qualified in this way. First, we shall be concerned only with
that function of the ceteris paribus condition that is necessitated by the
incompleteness of the starting-description. Thus, a ceteris paribus condition is
required in those rules of this kind that are justified by strong as well as weak
principles. But in the case of strong, as opposed to weak, principles the ceteris
paribus conditions could be dispensed with if the descriptions were in effect
made complete, by more precisely qualifying the main clause in ways which I
shall presently suggest. Rules derived from weak principles, however, would (p.
124) require a ceteris paribus condition in any case, since that condition would
be transmitted to the general judgement-rule from the weak grounding principle
itself, just as it is transmitted to particular judgements. In the following
discussion I shall ignore this second aspect or function of the ceteris paribus
condition, for I wish to suggest how the rules can be so qualified that a ceteris
pari-bus clause of the first kind is not required. In effect, then, I shall
beexplicating only rules that are justifiable by strong principles. The rules that
are justifiable by weak principles are not far to seek. To determine them, we
need merely add ceteris paribus conditions to rules that are justifiable by the
corresponding strong principles.
Secondly, the ceteris paribus clause is neither the only nor the most illuminating
type of qualification that can be added. I shall show how such a condition can be
displaced by more explicitly determinate ones.
Page 6 of 38
RULE-UTILITARIANISM
While on the topic of ceteris paribus rules it is worth noting that one
interpretation of this condition is inappropriate to rules justified by appeal to
utilitarian generalization. It has been suggested by Singer (pp. 99–100) that
moral rules have the form ‘A is generally (or usually, or probably, &c.) wrong (or
right).’Now this is a common enough form of moral rule; it seems suitable, for
example, for cautionary or summary rules. But such terms as ‘usually’,
‘probably’, and ‘generally’ are not appropriate in the present context. First, it
should be reaffirmed that ‘other things being equal’ is merely one possible
formulation of the ceteris paribus condition. We might alternatively employ
notions such as ‘prima facie … (e.g. wrong)’, ‘may be presumed … (e.g. wrong)’,
‘presumption … (e.g. against)’, ‘good reason … (e.g. against)’, and so on. For our
purposes, nothing hangs upon the choice within this range. But when
presumptions are established in the way I have suggested, on the basis of the
value of the tendency of a generic description in connexion with a form of
utilitarian generalization, we cannot suppose that such acts are generally (or
usually or probably) wrong. A presumption against lying, for example, is fully
compatible with most instances of lying, or lying in most kinds of circumstances,
not being wrong. We might say that presumptions as such have no quantitative
implications. Our understanding of the ceteris paribus qualifier (p.125) must
be in quantitatively neutral terms, such as is provided by the ‘other things being
equal’ formulation. Conversely, if we want to explicate rules of the form ‘A is
generally wrong’, we cannot do so by means of any straightforward application
of utilitarian generalization.
It would be misleading to leave the rules at this stage of analysis. They can be
explicated further, two means of which I shall now indicate.
Concise rules first. From our analysis of utilitarian generalization we know, not
only how presumptions can be established, but also how they can be rebutted.
From the specific dimensions of a given principle we can determine the exact
kind of grounds which can be employed in rebutting such a presumption. Such
grounds, which I shall call exception-making criteria, can be appended to the
ascriptive clause (in place of a ceteris paribus condition) to yield a concise (but
theoretically complete) rule.
Consider for example the simplest kind of case, the rule against lying that
presumably is justifiable by a (strong) negative non-comparative principle. The
concise form of such a rule is ‘Lying is wrong, except when its generalized utility
in a given case (in respect of a complete general utilitarian description) is not
negative.’ This rule tells us in schematic terms but none the less explicitly all
that is implied about lying by this form of utilitarian generalization. The new
qualification tells us what kinds of considerations are relevant and sufficient for
our purposes.
Page 7 of 38
RULE-UTILITARIANISM
Notice how such a rule is generated. We must first ascertain (we must know or
be justified in assuming) that lying has a negative generalized utility.
Consequently, whether such a rule against lying can actually be generated from
a form of utilitarian generalization depends upon matters of fact and our
evaluation of effects. Similarly, it remains to be seen what exceptions to such a
rule (i.e. qualifications upon the ascriptive clause or rebuttals of the
presumption) could be justified by a particular form of utilitarian generalization.
In general all we can say is that the exceptions will accord with those cases that
would be found not to be wrong by a direct application of utilitarian
generalization.
This type of rule suggests a reformulation of utilitarian generalization into

primitive rule-utilitarianism. We can view the (p.126) principles as schemata
for the generation of such primitive rules, along the following lines: ‘If the
tendency associated with a given description A is undesirable, then a rule of the
form “A is wrong, except … ” can be justified.’ The appropriateexception-making
criteria must be filled in, according as the principle in question is negative or
positive, comparative or non-comparative.
(Actually, this schema is insufficiently general: it fails for example to exhaust the
implications of the positive and comparative forms, for these types of principles
allow presumptions against acts which have indifferent or even desirable
tendencies. We can therefore do one of the following things. Taking a cue from
Harrison, in his rough draft of a form of utilitarian generalization (p. 114), we
might have three rule-schemata corresponding to three kinds of ascriptive rules.
The main clauses of such rules would express presumptions for, against, or
indifferent with respect to acts having some generic descriptions which are
associated with desirable, undesirable, or indifferent tendencies, respectively.
Thus, we could have ‘ … wrong, except … ’, ‘ … wrong not to do, except … ’, and
‘ … not wrong, except … ’ rules. The exception-making criteria to be appended
will vary accordingly. Alternatively, we might retain only the ‘ … wrong, except
… ’ form and require that presumptions be based, for example, upon relative
generalized utilities for the application of such principles as primitive rule-
utilitarianisms. This approach has the advantage of economy in the number of
kinds of rules, but has the disadvantage of excluding non-relative descriptions
such as ‘lying’—descriptions in which we are most usually interested. For
theoretical purposes the choice is a matter of detail. I shall continue to employ
as examples only cases of ‘ … wrong, except … ’ rules which are based upon non-
relative descriptions.)
Page 8 of 38
RULE-UTILITARIANISM
I want now to indicate the alternative schema for theoretically complete

primitive rules. Concise rules have the virtue of brevity; they tell us all we need
to know in a schematic way about a given kind of act. But this virtue has a high
price, for sometimes we want to know whether a given set of circumstances
constitutes an exceptional condition, i.e. a condition in which a presumption can
be rebutted. For the latter purpose we can employ expanded (p.127) rules.
Expanded rules are in effect developed out of concise ones: the implications of a
given concise rule are unpacked and become explicit in the corresponding
expanded rule.
What I have in mind is this. Instead of mentioning the exception-making criteria

in the qualifying clause, we can list all those conditions the satisfaction of which
in given cases suffices to rebut the presumption against the act as described in
the main clause of the rule. The qualifying clause becomes a complex disjunctive
clause, i.e. a disjunctive list of exempting-conditions. Thus, instead of appending
to the ascriptive clause a concise qualifier like ‘except when the generalized
utility … ’, we append the expanded qualifier, a clause of the form ‘except when
A is also B, or C, …, or K’. Here B, C, …, and K (which may be simple or complex
descriptions) are exempting-conditions.
For example, taking the simple case of a rule justifiable by a negative, non-
comparative principle, if A is presumed wrong on the basis of its associated
undesirable tendency, then B is an exempting-condition if the generalized utility
of acts AB is not negative. ^AL might be ‘lying’ and ‘2?’ could then be
‘performed in order to save another's life’.
(Here we find another technical complication which need not detain us. We
could require that exempting-conditions be so specified that ‘AB’ AC and so on,
represent complete relevant descriptions of acts. But this may be unnecessarily
stringent. We might allow ‘B’ to be included if, in virtue of the fact that an A is
also a B, the presumption against the act that is based uponits being an A is
rebutted. In other words, B would be a ‘right-making’ condition sufficient to
outweigh the ‘wrong-making’ condition A. But then such a rule might fail to deal
with other instances of A which are not made right simply because they are also
B. I am thinking of cases in which there is more than one wrong-making
condition, e.g. D as well as A, and wherein B is not sufficient to outweigh AD,
although BE may be sufficient. We then have a choice. We can include more
complex exempting-conditions in the rule so that it becomes, in part, ‘A is wrong,
except when A is also 5, …, or BDE,. … ’ Or we can construct an additional rule
beginning with AD, one of the exempting conditions of which would be BE, but
not B.)
Page 9 of 38
RULE-UTILITARIANISM
All the concise rules of a given primitive rule-utilitarianism (p.128) can in

principle be developed into expanded rules. But it is important to note what kind
of claim I am making in saying this. I am not saying that we can in practice
complete all the calculations required to expand a concise rule, that we can
complete the listing of all the exempting-conditions of all the expanded rules. I
am simply proposing a model for viewing the import of a given principle vis-a-vis
a given kind of act. The notion of an expanded rule is to be used as a device for
conceptually cataloguing presumptions and rebuttals. There is perfectly good
sense in claiming that such presumptions and rebuttals are implied by a given
principle (in a given context). And if there is sense in such a claim, then it may
be granted that these rule-schemata can be useful devices for analysing such
implications en masse.
I am especially interested in two classes of exempting-conditions that would be

included in those rules concerning acts that have associated thresholds. These
are (i) what I shall call maximizing-conditions: designed to maximize the utility of
a practice, to make the best of a generally good situation. These conditions are
based upon the existence of a (relatively) good-producing practice or the non-
existence of a (relatively) bad-producing practice; (2) minimizing-conditions:
designed to minimize the disutility of a practice, to make the best of a generally
bad situation. These are based upon the non-existence of a good-producing
practice or the existence of a bad-producing practice. These classes of
exceptions do not exhaust all the conditions which would probably be included in
the rules, but they have this significance: they are based upon the pattern of
others doing the same, where ‘the same’ represents a description in vacuo. (If
we strike maximizing-and minimizing-conditions from the rules they become
rules in vacuo.) These exempting-conditions are therefore crucial to the
equivalence of primitive rule-utilitarianism and simple utilitarianism.
I shall now try to indicate by means of an example the nature of maximizing-and

minimizing-conditions. It is important to be clear about these because they will
be central to our extensional analysis of other kinds (non-primitive kinds) of rule-
utilitarianism. Suppose that six of us are driving along in our jointly-owned
automobile. It is important for each of us—perhaps for all of us jointly—to get to
our destination as quickly as possible. But (p.129) owing to our limited
resources our car is not the newest and freshest on the road. Its paint is worn
and it has no tail-fins or shining hub-caps. But its main fault is that it does not
run well. It seems malevolently predisposed to stalling at inconvenient places,
e.g. upon ascending hills. And since the battery is weak it can be started again
only by brute force: by getting it moving and throwing it into gear.
Page 10 of 38
RULE-UTILITARIANISM
Suppose further that we know this car and its idiosyncrasies well. Now the car
stalls—this time, when we stop along a level stretch of road. Moreover, we have
only twenty yards or so to push the car before a descent, and thus we know that
we could all get it moving again just by pushing it and coasting it down the hill a
short way. But we are all tired, having had to do this sort of thing too often. It
will be hard work to start the car. None the less, it can be done. Indeed, we also
know that five of us could manage the job. (The threshold effect, getting the car
moving, & c, requires less than universal performance.) And whether five or six
of us do it, the reward in getting the car started and continuing on our way
would undoubtedly outweigh the several efforts that would be involved in doing
it. (No question but that everyone's pushing would yield good effects on the
whole: a good-producing practice.)
But since five could do the job without undue strain, and since (let us suppose)
the relief for one who does not push would outweigh the sum of extra efforts
required by the other five in pushing in such a case, a maximizing-condition is
satisfied. For the practice of pushing the car would have better consequences
over-all if one of us did not push. The good threshold effects would be produced
anyway, and the relief afforded by a general but not universal practice would
enhance the net-balance of good produced. Thus one exception could be made.
But a problem arises even in such a simplified example which does not arise in
the more complex cases such as voting where there can be assumed to be a
relation of anonymity among the several agents. For in the present case no one
is in a position to exploit the maximizing-condition. One of us could not, for
example, simply opt out of pushing for the sake of increasing the net-balance of
good produced. Jones wants to rest; but so do all of us; why he? It is possible
that more harm than good could (p.130) come out of one's insisting upon being
excused without a special reason beyond that of increasing the net-balance of
good. An argument might ensue, tempers could flare, and a generally bad time
would result.
Suppose, however, that others are pushing and that one merely feigns pushing.
Feigning also takes some effort, but it might sometimes yield an enhanced net-
balance of good. By means of the simple expedient of secrecy one could exploit
the maximizing-condition. But a natural objection against such behaviour is that
it is not fair. Someone would be ‘getting away with’ working less at the expense
of others. Even if the others did not have to push proportionately harder to make
up for Jones's failure to push, Jones's act might still be considered unfair in that
the benefits would be inequitably distributed.
Page 11 of 38
RULE-UTILITARIANISM
The issue of fairness is one to which we shall return. Here I wish merely to
acknowledge that the obvious unfairness of Jones's demand or behaviour can
itself have disutility. But it is not clear that in all such cases in which one acts
alone to exploit a maximizing-condition—even if one acts unfairly—there will be
a net-loss of utility. Nor is it clear that the notion of fairness always applies to
cases involving maximizing-conditions. If one abstains from voting because of
inconvenience when in fact enough others will vote to enable, say, the best man
to be elected, it is not at all clear that one's act is unfair (though it may be
wrong, although not on utilitarian grounds). Nor does it seem that abstaining
would result in a net-loss of utility, not only if one person abstains, but if all
others in a relevantly similar position do so as well.
The objection might be made that the primitive rule that would allow such
exceptions based upon maximizing-conditions is purportedly moral, and thus
must be publicizable. The objection would run this way: if there were a general
awareness of such an exempting-condition, if the import of the rule including
that condition were made public, it would be impossible to take advantage of the
exception. Everyone would try to act on the exception and thus no one would do
what was required. Or else the obvious unfairness of the act (exploiting the
maximizing-condition) would itself have bad effects.
I shall make just two comments on such an objection for the (p.131) present.
First, it is impossible for everyone covered by a rule to exploit a maximizing-
condition. If everyone tried to do so, the good-producing practice would collapse
(e.g. the car would not be pushed at all, or no one would vote). It is not enough
merely to think that one is acting upon a maximizing-condition. Whether one can
possibly do so depends upon the actual behaviour of others. Secondly, the
question whether a rule is publicizable must be distinguished from the question
whether it is in fact public. It is perfectly consistent and intelligible for a moral
theory to be such that its rules include exceptions based upon whether or not
others are generally conforming with, are aware of, or accept the rules. If the
theory is effectively publicized and perhaps accepted and acted upon, then some
of the exceptions which would be allowed under different circumstances will not
be allowed in these.
Page 12 of 38
RULE-UTILITARIANISM
Let us now consider a minimizing-condition, using the same example. Suppose

that three of the six of us who are travelling in the car refuse to push when it is
stalled. These three feel that they have had all the pushing they can stand and
simply refuse to do any more. They refuse to be convinced by argument; they
will not see—or act upon—their own and everyone's best interests. Obviously,
the silliest thing one could do in such a case, where five must push for any good
effect and only three are willing to do so, is push anyway. If one thought that in
pushing one acted upon a primitive rule (doing that which would have the best
effects if everyone did it), this would be a mistake. A primitive rule covering
such a case would include a minimizing-condition indicating in effect that it
would not be wrong not to push when the car would not get started anyway
(even though it would be wrong to fail to push when the requisite number were
willing and able).
Maximizing-and minimizing-conditions should be kept distinct. Although both

depend upon others doing or not doing the same, they depend upon such factors
in different ways and have different utilitarian points. As we shall see, although
maximizing-conditions are somewhat problematic—some would be rejected from
a non-utilitarian point of view—they are not generally excluded from the rules of
non-primitive rule-utilitarianisms. Minimizing-conditions, on the other hand,
which (p.132) seem eminently sensible, are not clearly included in the rules of
such theories. Thus, the non-primitive rule-utilitarianisms differ from primitive
rule-utilitarianism in the wrong ways, as I shall argue.
I should mention finally that what I am calling ‘minimizing-conditions’ seem to

correspond with those sometimes called ‘states of nature’ by proponents of
utilitarian generalization. (See, e.g., Harrison, pp. 125–9, and Singer, pp. 152–
61.) I find the term ‘state of nature’ inappropriate, however, first, because some
minimizing-conditions are satisfied in circumstances which are far too mild to fit
the catastrophic connotations of such a term as ‘state of nature’, and second,
because the circumstances which would satisfy some minimizing-conditions (e.g.
general lying, general promise-breaking) are so serious, so disastrous, as to
suggest a state of affairs which cannot intelligibly be supposed to obtain along
with ordinary human existence.
Page 13 of 38
RULE-UTILITARIANISM
Now some general comments on primitive rule-utilitarianism. In the first place,

such a rule-utilitarianism can in principle be developed. That is to say, general
judgements that are suitably and exhaustively qualified and that may be viewed
as moral rules can be justified by any given form of utilitarian generalization. I
am not suggesting that we can in practice grind out a set of rules completely and
exactly exhausting the implications of such a principle. But this is not because
such a set of rules could not be extensionally equivalent to the principle; the
difficulty is not a result of the transition to rules (i.e. to intermediary general
judgements that may be viewed as rules). It is essentially the difficulty we faced
before in considering the application of utilitarian generalization directly to acts:
the factors are so complex that we could not necessarily run through all of them
in practice.
It may be misleading to speak of a set of rules extensionally equivalent to one

such principle. For one thing, there might be several, even innumerable such
sets for a given principle, since there may be several, perhaps innumerable sets
of ascriptive clauses (starting-descriptions) that could be employed in the
construction of a set of rules completely exhausting the implications of a given
principle. Of course we are not in the least concerned with economy in such a
set—we are not concerned, for example, with a set containing the smallest
number of shortest (p.133) possible rules. We are not concerned primarily
because we have imposed no special conditions upon the rules of our primitive
rule-utilitarianisms. We are not supposing, for example, that these general
judgement-rules match any given set of conventionally accepted rules, e.g. the
moral code of a given community. These are not rules in the sense of actually
being accepted by the members of a given community; they are judgements that
might be viewed as rules and that are also implied by given principles.
Primitive rule-utilitarianisms must also be seen in the light of earlier

observations about consistency. If the sets of rules preserve extensional
equivalence with the grounding principles, there will be as much danger—but of
course no more danger—of inconsistencies in the sets of rules as among the
judgements derived directly. The comparative principles seem to be internally
consistent; therefore the rules based upon them will be mutually consistent. But
we would expect that the strong non-comparative principles would generate
mutually incompatible rules; these inconsistencies are localized in ‘lesser of two
evils’ and similar situations, when more than one act is counted as wrong not to
do, or when all the alternatives are counted as wrong. Again, the weak non-
comparative principles will necessarily yield mutually consistent rules: this is the
result of retaining the ceteris paribus condition which now functions only to
eliminate the inconsistencies in the corresponding strong non-comparative
principles.
Page 14 of 38
RULE-UTILITARIANISM
Most important of all, it follows, from the fact that such rule-utilitarian theories
can be developed extensionally equivalent to the forms of utilitarian
generalization, that an appeal to rules as such makes no effective difference in
the import of a utilitarian theory. For it is not a question what steps our
reasoning must take, nor what forms our judgements must assume, but simply
the content of the rules and the ultimate import of the principles. Therefore, if
rule-utilitarianism is to offer a substantive alternative to simple utilitarianism,
there must be special conditions imposed upon the rules so that they are not
unrestrictedly grounded upon generalized utility and so that they may therefore
diverge from the rules of primitive rule-utilitarianism.
As I have already suggested, however, rule-utilitarians do have an alternative—

indeed, an indefinite number of alternatives.
(p.134) These should not be viewed as modifications of primitive rule-

utilitarianism, as theories altered and patched up so as to guarantee a measure
of non-equivalence with, say, Act-Utilitarianism. For these non-primitive rule-
utilitarianisms have actually been developed in many different ways.
Nevertheless, it will be helpful in trying to understand the import of the
alternative theories to compare them extensionally with primitive rule-
utilitarianism. Thus, my explanation of how rules could be generated from the
forms of utilitarian generalization had a twofold purpose: first to explicate the
notion that utilitarian generalization can be viewed as the ground of moral rules,
and second to show that a mere appeal to rules does not necessarily yield non-
equivalence with simple utilitarianism. But my elaboration of the details, my
brief discussion of maximizing-and minimizing-conditions, was intended to lay
the basis for a systematic comparison of Act-Utilitarianism with its non-primitive
rule-utilitarian alternatives. This can be done by means of the primitive rules of
(GU), the general utilitarian analogue of Act-Utilitarianism. For our purposes we
may restrict our attention now to this form of primitive rule-utilitarianism, which
I shall call Primitive Rule-Utilitarianism (PRU).
Page 15 of 38
RULE-UTILITARIANISM
In order to make perfectly clear the sort of comparison I shall make I want to
propose that we view the several alternative theories in the following way. It
seems reasonable to suppose, first, that as far as action-guidance is concerned,
all the rules of most (perhaps all) alternative rule-utilitarian theories can be
translated into one limited normative vocabulary such as ours based upon the
term ‘wrong’. Regardless of whether the rules are normally in the ascriptive
mood or some other (e.g. prescriptive), they can all presumably be translated
into the former. For example, acts that on other theories ought not to be done, all
things considered, or that rules like ‘Don't do A’ tell us not to do, are therefore
wrong in the translation. Acts we ought to do, or we are told to do (′Do A are acts
it would be wrong not to do. Other acts are simply right. If these theories have
second-order rules to adjudicate conflicts among incomplete first-order rules,
their directions can presumably be translated down into more full-bodied and
complete first-order rules of the kind we have been examining.
(p.135) Moreover, it seems reasonable to suppose that, in many (perhaps all)

cases, the specific rules in the sets justified by these alternative theories can be
so generated as to maximally overlap with the set of primitive rules. Of course
we do not necessarily have just one set of primitive rules for (PRU), since there
may very well be an indefinite number of complete sets containing exactly the
implications of (GU), but which can be distinguished by the fact that they are
generated from different sets of starting-descriptions. This applies to other
theories as well: they may very well have several sets of rules that could equally
well be justified, implying exactly the same things but formulated in different
ways. All we require for our purposes, however, is that there be one set of
primitive rules upon which we can fix, which yields (or is based upon) one set of
starting-descriptions and ascriptive clauses.
Page 16 of 38
RULE-UTILITARIANISM
Now if we take two corresponding sets of rules, a complete set of primitive rules
on the one hand, and a complete set of rules for an alternative (non-primitive)
theory generated as far as possible by beginning with the same ascriptive
clauses as are employed in generating the primitive rules, then we may find two
kinds of differences between the sets. There may be rule-gaps, wherein a rule of
one theory based upon a certain starting-description is simply not included in
the corresponding set of rules for the alternative theory; and there will be
exception-gaps, wherein two corresponding rules with identical ascriptive
clauses have different sets of exceptions (i.e. different exception-making criteria
which yield differences in the expanded lists of exempting-conditions). We can
treat the former as a limiting case of the latter and restrict our attention to one
type of difference between the corresponding rules of the alternative theories:
exception-gaps. By noting general characteristics of such gaps between the
primitive rules of (PRU) and those of alternative rule-utilitarianisms we can
determine schematically the differences in import between these two kinds of
rule-utilitarianism. We shall see that this sort of contrast is most directly related
to those threshold considerations upon which the equivalence argument is based
and, among these, those to which the arguments from fairness are most
relevant.
(p.136) B. Non-Primitive Rule-Utilitarianism

In the following I shall not consider all the possible varieties of rule-
utilitarianism. I shall restrict my attention to the predominant type of theory
actually proposed, which has been offered as a utilitarian alternative to Act-
Utilitarianism. These non-primitive rule-utilitarianisms may be said to be
analogous to Act-Utilitarianism and to (GU)—its general utilitarian analogue—or
(PRU)—its primitive rule-utilitarian analogue—in the sense that in each case the
notion of maximizing utility is centrally placed. The rightness or wrongness of
acts turns in some way, directly or indirectly, upon their being conducive to the
production of the best effects.
The idiom commonly used in the formulation of rule-utilitarian theories is

somewhat different from that which we have been using. But the common,
central terms such as ‘right’ and ‘wrong’ have the same, more or less technical
sense: they are contradictories. For example, the theory with which we shall
primarily be concerned, sometimes called ‘Ideal Rule-Utilitarianism’, may
roughly be formulated in the following way:
(IRU) An act is right if, and only if, it conforms to a set of rules general
acceptance of which would maximize utility.
Page 17 of 38
RULE-UTILITARIANISM
An act conforms to a set of rules only if the rules do not imply that the act is
wrong; or, in the terminology of prescriptive rules, only if they do not indicate
that the act should not (ought not to) be done, all things considered. I shall deal
with the notion of acceptance presently. Suffice it to say for now that a rule
isaccepted only if one tries to apply and act upon it. (See R. B. Brandt, Ethical
Theory (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1959), pp. 396–400,
and, by the same author, ‘Toward a Credible Form of Utilitarianism’, in H.-N.
Castaiieda and G. Nakhnikian, Morality and the Language of Conduct (Detroit:
Wayne State University Press, 1963), pp. 107–40, for formulations and
development of theories of this type.)
For the sake of contrast, I shall suggest how (AU) and (GU) would look in the
characteristic form in which rule-utilitarian principles are rendered. Act-
Utilitarianism becomes:
(AU) An act is right if, and only if, its total effects are no worse than those
of any alternative.
(p.137) Similarly, its general utilitarian analogue becomes:
(GU) An act is right if, and only if, its completely relevantly determined
tendency is no worse than that of any alternative.
Now let us try to recast (GU) into explicitly primitive rule-utilitarian terms. We
discover an interesting and illuminating difficulty. Consider the following
principle:
(SRU) An act is right if, and only if, it conforms to a set of rules general
conformity to which would maximize utility.
Brandt calls this a ‘specious’ rule-utilitarianism (‘Toward a Credible Form of

Utilitarianism’, pp. 119–23) because ‘it has identically the same consequences
for behaviour as does act-utilitarianism’ (pp. 120–1). Brandt evidently is claiming
what I would claim by saying that (AU) and (SRU) are extensionally equivalent.
But these principles are not equivalent, and that is because (SRU) is not the
rule-utilitarian rendering of (GU). I shall try to show this by reference to
maximizing-and minimizing-conditions. The differences between (AU) and GU),
on the one hand, and (SRU), on the other, can be shown in relation to systematic
differences between the primitive rules of (PRU)—which is extensionally
equivalent to (AU) and to (GU)—and the ‘specious’ rules of (SRU).
Page 18 of 38
RULE-UTILITARIANISM
The difference is briefly this. Whereas minimizing-conditions can and must be

included in the primitive rules, their inclusion in the specious rules is
indeterminate. Consider, for example, only those primitive rules against acts A
which are based primarily upon the undesirable threshold effects which could be
produced by the general performance of A. Minimizing-conditions allow for
circumstances in which there is in fact a general practice of A (and thus A bad
threshold effects are produced): acts A would be allowed under such
circumstances. That is, in so far as the bad effects of acts A are threshold-
related, if the threshold is already passed, then further acts A would not be
considered wrong. Minimizing-conditions, as I have said, are designed to make
the best of such a generally bad situation.
But to say that minimizing-conditions allowing acts A are based upon the general
performance of A is to say, in effect, (p.138) that they are based upon a general
failure to conform to the rule against A.For if the rule against A were generally
conformed to, then too few acts A would be performed to allow the production of
the undesirable threshold effects. (Some acts A might be performed when the
rule against^ is generally conformed to, for some such acts might be allowed by
maximizing-conditions. But when A's undesirable threshold effects are produced,
it is not possiblethat everyone, or most people, are conforming to the rule
against A. It is not possible that the rule is generally conformed to.)
Now the difference between Act-Utilitarianism and (PRU), on the one hand, and
(SRU), on the other, turns upon the fact that in the latter case a general
conformity test is employed in justifying rules. In determining a set of rules for
(SRU), we ask what set of rules would have the best consequences if they were
generally conformed to. But if we take two alternative sets of rules, one of which
does and the other of which does not include minimizing-conditions, but that are
identical in every other respect, then the general conformity test will not
distinguish between these two sets. For the test is based upon the supposition of
general conformity. Under such a supposition, the minimizing-conditions will
simply be vacuous. For the minimizing-conditions are satisfied only when there
is not general conformity. Thus, no difference will show up between the rules
that contain minimizing-conditions and the rules that do not, under the general
conformity test. This means that (SRU) may be regarded as indeterminate for
this class of cases, that a set of rules without minimizing-conditions is
compatible with (SRU). And thus it cannot be claimed that the implications of
(SRU) and those of (AU) are identical, for (AU) is by no means indeterminate for
this class of cases.
Page 19 of 38
RULE-UTILITARIANISM
The point can be put in another way. It is misleading to analyse (PRU) in

‘conformity’-terms parallel to the terms appropriate to (SRU). (PRU) is designed
to cover all cases from the point of view of the agent's act or choice of action.
One of the factors which is given full rein in determining the rightness or
wrongness of an act under (PRU) is whether others are not, for example, doing A
when there is a general rule against A. If others are generally doing A then the
best possible consequences on the whole cannot be produced, in the sense that
better (p.139) consequences would be (or would have been) produced if fewer
did A.But given these circumstances wherein A is generally performed, the
agent's act is assessed according to the best possible consequences that can
then be produced. A principle like (AU) or (PRU) is so designed that, if it were
conformed to, or correctly applied and acted upon, in any particular set of
circumstances, then the best possible consequences that could be produced
under those circumstances would be produced. But the general conformity test
employed in applying (SRU) does not provide for all such contingencies. It
provides only for conditions in which the rules are generally conformed to. Thus,
the implications of (AU) and (PRU), on the one hand, and of (SRU), on the other,
are identical only when these principles are in fact generally conformed to. And
this is a very special case indeed.
How then are we correctly to formulate (PRU)? Perhaps this way:
(PRU) An act is right if, and only if, it conforms to a set of rules conformity
to which in the case in question would maximize utility.
It is understood, of course, that the case in question is always to be treated as

one among a number of similar cases, i.e. completely relevantly determined acts,
classified on general utilitarian grounds.
It is clear why Brandt supposed that (AU) and (SRU) are extensionally
equivalent. He was aware that others' behaviour is a circumstance relevant to
the efficacy of acts and to their descriptions, and that this factor must, in
consistency, be taken into account. (SeeBrandt, ‘Toward a Credible Form of
Utilitarianism’, pp. 121–3.) Moreover, there is no reason for supposing that
(SRU) and (PRU) would differ in respect of maximizing-conditions. For, as I have
already indicated, general conformity to a rule against A is compatible with some
people doing A, provided that A is not generally performed. Brandt failed to
distinguish between two radically different kinds of circumstances in which
others doing the same must be considered: he failed, in effect, to distinguish
minimizing-from maximizing-conditions.
Page 20 of 38
RULE-UTILITARIANISM
(p.140) A comparison of (PRU), and thereby of (AU), with Ideal Rule-

Utilitarianism can be developed along similar lines. In this case, however, the
comparison must be rough, for the ideal rules of (IRU) are determined by a
general acceptance test which involves more factors than those involved in a
general conformity test. A word is in order on the relation of rightness and
wrongness to the maximization of utility under (IRU).
Ideal Rule-Utilitarianism (or a paradigm principle of this type) is formulated,

once more, as follows:
(IRU) An act is right if, and only if, it conforms to a set of rules general
acceptance of which would maximize utility.
On the Ideal Rule-Utilitarian approach, acts are not assessed on the basis of
their own simple or generalized utilities. The rules themselves are assessed as
functions of the utility of their general acceptance—their acceptance-utilities,
that is, the value of the effects of the general acceptance of the set of rules. Acts
derive their rightness or wrongness from what the validated rules have to say
about them.
Now in what does such general acceptance consist? Roughly, it would seem that
one accepts a given rule or principle when one adopts and avows or at least
makes a reasonable effort to apply and act upon it. This entails some measure of
success in doing what the rule or principle requires, that is, in doing right acts
only. Not perfect conformity: success cannot be expected in every case. For one
can mistakenly infer from a rule or principle things that the rule or principle
does not imply. Some allowance must therefore be made for mistakes, and also
for failures to do what one concludes, in any case, one ought to do. Thus general
acceptance differs from general conformity. In the former case a great deal of
conformity is presupposed, but supposedly somewhat less than in the latter—
although it is not clear how much less. Thus the ideal rules will not necessarily
be the same as the rules of (SRU), since differences in conformity can have
significant effects upon the utility of a given rule or set of rules. In other words,
the conformance-utilities and acceptance-utilities of given sets of rules can differ
markedly. And since this is so, the set having the highest conformance-utility will
not necessarily be the same as the set having the highest acceptance-utility.
(p.141) But in any event, if we judged sets of rules on the basis of the utility
that could be gotten from their general acceptance, we would find that two sets
of rules, that are exactly similar in all respects except that one set includes
minimizing-conditions while the other does not, would not be differentiated. For
again, since we are postulating general acceptance in the general-acceptance
test, the minimizing-conditions would in general be vacuous. Thus general
acceptance of either set would, roughly speaking, have the same consequences.
Page 21 of 38
RULE-UTILITARIANISM
There is a further reason for supposing in the case of (IRU), although perhaps
not in the case of (SRU), that general acceptance requires the omission of
minimizing-conditions, and that (IRU) is not merely indeterminate for this class
of cases. The principle we are examining is called ‘Ideal Rule-Utilitarianism’
because the rules are supposed to amount to ‘ideal rules’ (or, as Brandt calls
them, ‘ideal prescriptions’) for the agent's community. They constitute a kind of
utilitarian social policy. (See Brandt, ‘Toward a Credible Form of Utilitarianism’,
p. 109 and passim.) They are the rules that, supposedly on utilitarian grounds, it
would be best if everyone tried to conform to. They are supposedly suitable for
general adoption as moral rules in the given community. Thus there is the image
of a public moral code, promulgated, codified. And when one begins to think in
such quasi-legal terms, it becomes impossible to allow that minimizing-
conditions can be included in the ideal rules. For example, when we frame a law
we frame it supposing that it will be generally observed; we do nqt and cannot
allow exceptions merely on the ground that it is not generally observed.
(Although they may become and may be officially recognized as ‘dead letters’,
laws do not incorporate minimizing-conditions excusing non-conformity simply
because they are not generally observed.) Moreover, since the consequences
would in fact be better if no minimizing-conditions ever were satisfied, perhaps
in drafting such an ideal moral code one would not entertain exempting-
conditions which are based upon a generally bad (non-conformist) state of
affairs.
Thus, we can say in general that Ideal Rule-Utilitarianism does indeed provide a
substantive alternative to Act-Utilitarianism. But it is not clearly the sort of
theory with which utilitarians can be satisfied. For to act upon a theory that does
not allow (or is (p.142) indeterminate with respect to) minimizing-conditions is
possibly to act, from a utilitarian point of view, in a self-defeating manner. An
argument could be developed along the following lines, similar to a type of
argument which we shall examine in greater detail presently. (1) The point of
(IRU) is to maximize utility. (2) But there are some cases in which applying and
acting upon (IRU) will not yield the best possible consequences, that is, when
(IRU) is not generally accepted. (3) Thus (IRU) cannot be an acceptable moral
principle, since in some cases (that is, when it is not generally accepted—which
is the normal state of affairs) applying and acting upon (IRU) would be self-
defeating.
Page 22 of 38
RULE-UTILITARIANISM
Rule-Utilitarians have been aware of this problem and sensitive to criticisms

along these lines. (See, e.g., Brandt, Ethical Theory, p. 400.) The consensus
seems to be that minimizing-conditions must, in effect, be acknowledged. But to
say this is to say one of the following: (1) Exceptions must be allowed to the
ideal rules that are not allowed by the governing principle. This is manifest
inconsistency. (2) The ideal rules are (let us suppose) indeterminate for the
range of cases in question. Thus the governing principle is unable to account for
an important range of cases. This makes the Ideal Rule-Utilitarian position
incoherent in a special way. For, since two differing sets of rules can be validated
by the general acceptance test, that is, rules that do and rules that do not allow
certain classes of exceptions, it appears that according to (IRU) some acts are
both right and wrong or neither right nor wrong. The former result is untenable;
the latter is in discord with the apparent generality of (IRU), the formulation of
which suggests that all acts are either right or wrong.
Brandt has given considerable attention in his recent work to the problem of
determining a set of rules ‘for an imperfect society’. (‘Toward a Credible Form of
Utilitarianism’, pp. 126–30.) He acknowledges, for reasons which overlap with
but are somewhat different from mine, that Ideal Rule-Utilitarianism ‘savors a
bit of the Utopian’ (p. 127). He proposes, consequently, a modification of the
latter theory that incorporates into the test for the set of rules a consideration of
the existing (de facto) moral beliefs within a given community. The resulting,
more ‘credible’ form of rule-utilitarianism, which I wish to comment upon briefly,
is presented as follows:
(p.143) The compromise I propose is this: that the test whether an act is
right is whether it is compatible with that set of rules which, were it to
replace the moral commitments of members of the actual society at the
time, except where there are already fairly decided moral convictions,
would maximize utility, (p. 129.)
The resulting principle is:
(CRU) An act is right if and only if it conforms with that learnable set of
rules, the recognition of which as morally binding, roughly at the time of
the act, by all actual people insofar as these rules are not incompatible
with existing fairly decided moral commitments, would maximize intrinsic
value, (p. 129.)
Page 23 of 38
RULE-UTILITARIANISM
It should be observed, however, that the de facto morality of a given community

can be grotesquely immoral. Thus, the resulting standards gotten by applying
(CRU) might be far too low. On the other side we find that one of the strongest
features of moral beliefs in societies like ours is a sensitivity to unfairness and
injustice. Consequently, the validated set of rules for such a society will be
determined, in part, not merely by considerations of utility, but also by
considerations of justice and fairness. On such a theory as (CRU), therefore,
non-utilitarian factors are introduced indirectly.
These observations suggest that the rule-utilitarian is gradually forced to forsake

a pure utilitarianism for a composite theory. In the next chapter I shall deal in
greater detail with some of the arguments from justice and fairness that are
particularly relevant to rule-utilitarianism and to utilitarian generalization—
arguments which tend to show that a pure utilitarian theory, whether or not it
involves rules, cannot be adequate. But first I want to examine the more
immediate question, whether principles like Ideal Rule-Utilitarianism can
reasonably be regarded as utilitarian. To this we shall now turn.
C. Utility and Rules

Because non-primitive rule-utilitarianisms differ substantively from Act-
Utilitarianism we can reasonably ask a question which might well have been
asked before, but the raising of which was made unnecessary by the arguments
for extensional equivalence.
(p.144) The question is: how is it possible for a utilitarian consistently to

embrace a theory that is, or is supposed to be, a substantive alternative to Act-
Utilitarianism? For whenever a person successfully applies and acts upon (AU)
he performs an act that will produce as good consequences as could possibly be
produced under the circumstances. Thus, to the extent that, say, (IRU) diverges
from (AU), there will be some cases wherein, if (IRU) is successfully applied and
acted upon, the best possible consequences that could be produced will not be
produced, or at the very least one has restricted one's choice of action by
regarding as wrong acts whose effects would not be worse than those of any
alternative.
A strictly analogous comparison could be made between non-primitive rule-

utilitarianisms and (GU) by speaking of the results of everyone's doing the
relevantly similar sort of thing. But the comparison could be developed more
sharply in terms of exception-gaps between primitive and non-primitive rules.
Wherever there is an exception-gap between the primitive rules of (PRU) and the
ideal rules of (IRU), that signals a case in which successful following of the ideal
rather than the primitive rule would forsake utility in the above sense. Thus it
would seem that adherence by an avowed utilitarian to a non-primitive rule-
utilitarianism such as (IRU) is inconsistent and therefore demands an
explanation.
Page 24 of 38
RULE-UTILITARIANISM
I shall now try to reconstruct one particularly relevant kind of argument by

means of which utilitarians may—and apparently sometimes do—justify their
rejection of Act-Utilitarianism and their adoption of an alternative ‘utilitarian’
theory. We can begin with a seemingly straightforward argument against Act-
Utilitarianism. It is sometimes claimed (and must be admitted) that we hardly
ever have all the information required for the application of (AU), that is, enough
information to warrant our assurance that what we infer about the rightness or
wrongness of acts is actually implied by the principle. Moreover, time presses
and we are often forced to act in haste, sometimes without adequate
deliberation. We are also careless and prone to error in favour of our own
interests. And sometimes we fail to do what we believe to be right—what we
believe to be required or allowed by our moral principles. Thus, from the fact
that (AU) is applied and (p.145) acted upon (allowing a reasonable margin for
failure), it does not follow that the best possible consequences are produced. To
try is not necessarily to succeed.
What does this argument prove? I want to distinguish two ways in which such
considerations may be employed. For the Act-Utilitarian they assume the
significance of practical obstacles to the successful application of his theory. For
the rule-utilitarian they may take on a radically different significance: they
suggest an argument which, if successful, strikes at the very core of Act-
Utilitarianism.
Consider the argument in practical terms first. It is admitted that mere

acceptance of Act-Utilitarianism does not maximize utility, for the reasons given
above. Since it may also be argued that our chances of success in maximizing
utility would be enhanced by our application of a simple set of rules (instead of
applying the Act-Utilitarian criterion directly), this may be turned into an
argument for ‘rule-utilitarianism’. That is, if we want to succeed in producing the
best consequences—the ‘point’ of Act-Utilitarianism—then we should not
calculate utilities directly, but should instead rely upon and appeal to a simple
set of rules.
But interestingly enough, the Act-Utilitarian is apt to accept this argument, to

use it for urging that in practice we observe a set of moral rules. He is apt to
recommend that we appeal to cautionary rules, rules that can be employed as
practical aids in determining the best course of action, rules that are more
reliable in their action-guidance than we are likely to be in our deliberations.
In order to understand the Act-Utilitarian's position, however, in order to see

how misleading it would be to characterize his employment of cautionary rules
as ‘rule-utilitarian’, it is necessary to draw a further distinction among types of
rules. This distinction, between theoretical (theory-dependent) and de facto
(conventional) rules, concerns what might be called theirstatus; it is a distinction
cutting across those we have already made which turn mainly upon form.
Page 25 of 38
RULE-UTILITARIANISM
Some rules owe their very existence to the fact that they are acknowledged,
observed, avowed, referred to—that individuals can and sometimes do appeal to
them in criticizing others' behaviour and in justifying or criticizing their own
conduct. We (p.146) determine that such rules exist and what they imply (what
their content or import is) by seeing how people behave and how they talk and
argue about their behaviour. These are de facto rules. This class includes, for
example, legal rules, traditions, conventions, and mores. It may be difficult in
practice to determine which de facto rules belong to a given group or society.
Legal rules are generally easiest to pin down since they are normally recorded.
Unwritten rules are, however, harder to verify. And in the realm of unwritten
rules—including conventional or de facto morality in general—we face the
difficulty of interlockingsub-groups in a particular society, each with its own
complex set of rules. It is also true that the content of some of the rules of most
groups is hard to determine—or indeterminate to some extent—because of a lack
of unanimity among the members of the groups. And some rules receive more
lip-service than conscientious application. Nevertheless, despite the difficulties
there are in determining which ones exist and just what they are, some such
rules exist in virtue of the status we accord them in our behaviour, loyalties, and
deliberations. And clearly it is of the very essence of such rules that they be—
within the group whose rules they are—public.
But the rules which we have been constructing in this study are theory-
dependent or theoretical. These rules are implied by a given theory. The rules
implied by a moral theory are purportedly moral, but they require no publicity or
general acceptance. They need not be generally observed, acknowledged,
accepted, or referred to. They need only be implied or justified by a given moral
theory; and thus they are dependent upon that theory; as theoretical rules they
are dependent for their correctness upon the validity of the theory.
Some theoretical and de facto moral rules can of course substantially agree; the
distinction I am making does not affect that possibility. The theoretical rule
against lying implied by a given moral theory, for example, might have exactly
the same content as a de facto rule against lying. That is, a given theory might
agree with current moral beliefs regarding which cases of lying are right and
which are wrong. Indeed, this kind of convergence is to some extent the aim of
moral theorists. But the two types of rules must none the less be distinguished.
That is to say, the grounds (p.147) upon which, according to one rule or the
other, an act is right or wrong are different.
Page 26 of 38
RULE-UTILITARIANISM
If we fail to keep this distinction in mind confusions arise. For example,

philosophers sometimes ambiguously speak of4 justifying’ or ‘accounting for’
moral rules. The rules to which the proponent of a given theory is committed, as
a proponent of that theory, are theoretical rules, e.g. primitive rules, ideal rules,
and so on. But the rules with which we are all familiar as part of conventional
morality are de facto rules, e.g. those the Act-Utilitarian recommends as
practical aids, as rules of thumb, as cautionary rules. Moreover, the de facto
moral rules of a given society cannot obviously be matched by the theoretical
rules of the kinds of principles we have been examining. For these theoretical
rules are, in principle, completely determinate (open-texture of terms aside). In
contrast, our conventional moral rules normally include a ceteris paribus
condition which cannot be analysed away into a determinate qualifier. There is
no de facto consensus regarding how conflicts among such rules are to be
settled. Confusion can be caused by the fact that ceteris paribus conditions are
sometimes incorporated into theoretical rules. But in the case of rules derived
from principles of the types we have examined, if a rule derives from a strong
principle then the ceteris paribus qualifier is eliminable; if it derives from a
weakprinciple, then how it can be eliminated—if it can be—depends upon the
remainder of the moral theory in question. Even if the moral theory is left
somewhat indeterminate, the question still arises whether the particular
theoretical and normal de facto uses of the ceteris paribus condition are exactly
equivalent. Only if the ceteris paribus condition of rules within a given moral
theory is expressly designed to be equivalent to conventional ceteris paribus
conditions is there a possibility of ‘justifying’ or ‘accounting for’ de facto moral
rules in the sense of deriving or justifying equivalents from the particular theory.
But if, for example, the ceteris paribus condition of de facto moral rules connotes
‘generally’, or ‘usually’, or ‘probably’, instead of ‘other things being equal’ as I
have developed it, then there will inevitably be a gap between derived
theoretical rules developed from theories such as we have examined and de
facto ones.
A second source of confusion is the attempt by some theorists (p.148) to

provide what may be understood as a theoretical explication of certain de facto
rules. This is essentially the development of general judgements beginning with
selected descriptions such as we have already discussed. Thus the Act-Utilitarian
may take the de facto rule against killing and transform it into ‘Killing is wrong,
except when the effects of killing would be better than the effects of not killing,
all things considered.’ And this is of course no longer our de facto rule against
killing, which is not at all generally understood to have such an Act-Utilitarian
explication. It is another question entirely whether this is a better rule, whether
it is one which we ought to observe—that is, whether it is the really moral rule,
in contrast with our merely conventional one.
Page 27 of 38
RULE-UTILITARIANISM
A third possible source of confusion is that, within some theories which have
been regarded as ‘rule-utilitarian’, the de facto rules of a given society are more
or less accorded theoreticalstatus. Thus in Toulmin's theory, for example, acts
are right or wrong according as they are so regarded under the moral code,
beliefs, and conventions of one's community. But Toulmin's use of de facto rules
should be distinguished from the Act-Utilitarian's, at least from that Act-
Utilitarian position I am now presenting. Whereas the Act-Utilitarian regards
them as rules of thumb that one would be well-advised to consult, on Toulmin's
account we are logically obliged to consult them—they are logically
indispensable.
In response to those practical considerations showing that there is a risk in

attempting to appeal to the Act-Utilitarian principle directly, the Act-Utilitarian
may recommend those ready-at-hand, relatively simple, conventional rules of
‘common-sense’ morality—de facto moral rules. These are supposed to be
reliable guides to right action: they presumably represent and summarize or
generalize upon centuries of mankind's experience regarding the usual utility of
socially important types of acts. The best means of ensuring that we do what we
must want to do as utilitarians, i.e. maximize utility, is to observe these de facto
moral rules.
But we must also adopt this escape clause: follow the rules, indeed, but not
when you know or are quite certain that breaking one will have better effects on
the whole than keeping to it. This (p.149) escape clause is essential to the Act-
Utilitarian's recommendation, and its inclusion clarifies the status of the rules.
For these rules are not determinants of rightness and wrongness; they are rules
of thumb, practical aids. The ultimate court of appeal is the Act-Utilitarian
principle itself: acts are right or wrong according to their relative simple
utilities. This point may be obscured by the doubt some utilitarians feel about
whether the escape clause can ever reasonably be supposed to be satisfied; for
the weight of the accumulated experience of mankind as embodied in the rules
is considerable vis-a-vis the chance of error, ignorance, and temptation affecting
our direct calculations of utility. In other words, some Act-Utilitarians hold that
rarely, if ever, are we practically justified in breaking the rules—that we would
be taking too great a risk in doing so. Thus the Act-Utilitarian must admit that
we may sometimes be unjustified in doing the right act when this involves
breaking rules that we have insufficient reason for breaking. Such a position can
be bolstered by drawing a sharp line between questions of rightness and
wrongness, on the one hand, and questions of praise or blame and responsibility
on the other.
Page 28 of 38
RULE-UTILITARIANISM
But it must be emphasized that such a practical addendum to the Act-

Utilitarian's position need not represent a modification of the Act-Utilitarian
principle itself. In recommending general observance of the de facto moral rules
the Act-Utilitarian is merely recommending a way of deliberating and of
effectively solving problems of conduct to best ensure conformity to the Act-
Utilitarian principle. And thus his elaborated position need not be viewed as
hypocritical or paternalistic. He is not committed to a dichotomy of moralities—
one, for the learned, philosophical, conscientious, trustworthy, and another, for
the ‘vulgar’; nor is he committed to keeping his moral beliefs in any way secret.
It is open to him to publish his full position, including his practical reasons for
recommending de facto morality, and to press hard upon whatever features of it
need emphasis.
The Act-Utilitarian's aim in recommending the adoption and observance of de

facto rules is to provide the most practical approach towards maximizing utility,
an approach that could reasonably be expected to have some success, not in
another world or at another time, but here and now. The Act-Utilitarian (p.150)
can claim of the rules he recommends what cannot be claimed for any other set
in a given community, not even for a set that an enterprising rule-utilitarian
might draw up as having, say, the highest acceptance-utility. We have these de
facto rules already in a way in which we cannot have any other set of rules.
But it should also be observed that the rule-utilitarian does not want merely to
recommend rules for adoption in the same way. The rules to which he is
committed have a different relation to his principle, to the criterion of rightness
and wrongness he accepts, to the notion of maximizing utility, than have the de
facto rules viewed as rules of thumb. The de facto rules are broughtin as
practical guides which (it is claimed) roughly summarize what the Act-Utilitarian
principle itself has to say about rightness and wrongness. But the rules with
which the rule-utilitarian is concerned are not practical guides, they are
themselves determinants of rightness and wrongness. That is to say, one cannot
determine the rightness or wrongness of particular acts (on the rule-utilitarian's
theory) without consulting the rules themselves. For as we have seen, on such a
theory as Ideal Rule-Utilitarianism, a set of rules is judged by its acceptance-
utility; and the rightness or wrongness of acts depends upon what the set of
rules with the highest acceptance-utility happens to say about them.
Page 29 of 38
RULE-UTILITARIANISM
Primitive Rule-Utilitarianism is, as it were, half-way between Act-Utilitarianism

and Ideal Rule-Utilitarianism in this respect. For the primitive rules may be
regarded as determinants of the rightness and wrongness of acts, but they are
not indispensable. Although the primitive rules are theory-dependent, what they
imply about particular acts can be gotten directly from the general utilitarian
principle from which (PRU) is developed. In the case of Ideal Rule-Utilitarianism,
however, appeal to the basic governing principle, (IRU), yields only directions for
the general acceptance test, and no criterion for directly assessing particular
acts.
Now let us consider how the rule-utilitarian may be regarded as exploiting the
practical obstacles to maximizing utility faced by (AU). The argument I shall
reconstruct cannot be ascribed to any particular moral philosopher, although
various versions (p.151) of it: are suggested in more than one paper on the
subject and in current discussions.
The argument against Act-Utilitarianism is sometimes put in this way: (i) The
point of Act-Utilitarianism is to maximize utility. (2) But if everyone applied and
acted upon (AU), utility would none the less not be maximized. That is, because
of the practical obstacles discussed above, some bad consequences would result
from everyone's applying and acting upon (AU)—or at least some good
consequences that could be produced would not be produced. (3) Therefore Act-
Utilitarianism is unacceptable as a moral principle, since its point would be
defeated (frustrated) by its general practice. The last point is sometimes made
by saying that (AU) is self-defeating or self-frustrating.
Let us call this the general coherence argument. I have purposely formulated it
in somewhat ambiguous terms (as it often is found) with the intention of
contrasting two interpretations. One is mistaken, the other is plausible. The
import and limitations of the latter can best be grasped by a brief examination of
the former.
But in order to develop this contrast I must first crystallize a distinction implicit
in the foregoing discussion, which will become essential as we proceed. The
distinction is between accepting and what I shall call following a rule or
principle. Acceptance, as I have said, suggests that one tries to apply and act
upon a given rule or principle correctly. But one follows a rule or principle only
when one succeeds in correctly applying and acting upon it, i.e. by doing what
the rule or principle requires.
Now consider the general coherence argument in its stronger form, in terms of
‘following’: (1) The point of Act-Utilitarianism is to maximize utility. (2) But if
everyone followed (AU), utility would not be maximized. (3) Therefore (AU) is
unacceptable as a moral principle, since its point would be defeated by its being
generally followed.
Page 30 of 38
RULE-UTILITARIANISM
This is not, however, an appropriate rendering of the general coherence

argument. For the argument is intended to make an issue of the difficulties
preventing successful acceptance of (AU), while in this strong version successful
acceptance, i.e. following, (p.152) is assumed. (2) is necessarily false. But this
version is the one sometimes suggested in the literature. Harrod (pp. 147–8) and
more clearly Harrison (pp. 131–4) seem to hold that the argument in terms of
‘following’ would work against (AU) but not against its general utilitarian
analogue (GU). For they suppose that the Act-Utilitarian is prevented from
taking threshold factors into account, that the Act-Utilitarian cannot
acknowledge effects that depend for their production upon general practices,
since he is committed to viewing and assessing acts separately, one at a time.
But this is of course that central error, that misapprehension about simple vis-a-
vis general utilitarianism which I have tried to expose.
The general coherence argument becomes more credible when we displace

‘follow’ by ‘accept’: (1) The point of Act-Utilitarianism is to maximize utility. (2)
But if everyone accepted, i.e. tried to apply and act upon, (AU), utility would not
be maximized. (3) Therefore (AU) is unacceptable as a moral principle, since its
point would be defeated by its general acceptance. The second premiss in this
case is plausible, for reasons already mentioned. It is this mode of argument,
based upon what we might call the relative acceptance-utilities of moral
principles, which seems to openthe possibility of fundamental criticism of Act-
Utilitarianism and the possibility of a more adequate, alternative utilitarian
theory. But there is much in it that requires clarification.
The general coherence argument in its most plausible interpretation can I think
be characterized as formal, self-referential, and comparative. It is formal in the
sense of not employing a substantive criterion on the basis of which principles
are to be judged; self-referential in the sense that each principle to which the
general coherence test applies is assessed according to one of its own features,
namely, its point; and comparative in the sense that a principle is evaluated in
comparison with alternative principles.
Consider the notion of formality first. Clearly this is not a consistency test; it is
not in that sense formal. Indeed, no one—so far as I know—has claimed that Act-
Utilitarianism is self-contradictory. Nor would it seem to be inconsistent in the
sense of implying a limited range of incompatible judgements, for its strength is
combined with a comparative feature.
Page 31 of 38
RULE-UTILITARIANISM
(p.153) It is not uncommon for philosophers to propose necessary, presumably

non-substantive tests for moral principles, rules, or maxims. Another condition
often suggested, for example, is that moral principles must not include essential
reference to particular persons, places, or things. I shall not deal generally with
the concept of such a formal test, but I hope that the following observations will
shed light upon the limitations of the general coherence argument. Apart from
the fact that it does not promise to work simultaneously against Act-
Utilitarianism and for some non-primitive form of rule-utilitarianism, it does not
seem to be a necessary test in the sense that any acceptable moral principle
must pass it.
I want to emphasize that the general coherence argument is formal because it is

too easy to slip into a methodologically questionable and viciously circular mode
of argument, one which has never clearly been distinguished from the general
coherence argument in discussions comparing the forms of utilitarianism.
Circularity is a danger primarily because the argument is normally employed by
utilitarians intramurally, as part of intra-utilitarian disputes, such as when the
rule-utilitarian seeks to show the superiority of his kind of principle. In this
context, a careless formulation of the argument may suggest that the criterion
according to which principles are to be assessed is utilitarian. That is, it may
seem that the argument rests upon a suppressed, higher-order criterion to the
effect that principles must be compared on the basis of their relative
acceptance-utilities.
If we understand the general coherence argument in this way, however, we seem

to be committed to holding that the principles assessed cannot possibly be
unjustifiable, ultimate moral principles. For if Ideal Rule-Utilitarianism or Act-
Utilitarianism or any other principle rests upon some higher-order utilitarian
principle concerning the relative acceptance-utilities of principles, then these
principles of right conduct such as (AU) and (IRU) become derivative and
secondary. And yet it is normally assumed by philosophers that, if any of these
principles are valid, they are ultimate and unjustifiable. They may be supported
by argument, but they cannot be derived from some other principles nor justified
by reference to them.
Page 32 of 38
RULE-UTILITARIANISM
(p.154) Moreover, even within the limited context of utilitarianism, the

argument tends to beg the question for such a theory as Ideal Rule-
Utilitarianism because it is a general coherence test. The condition supposed is
the general acceptance of a given principle. But as we have seen, the Act-
Utilitarian can offer a similar, but wow-general coherence argument. For Ideal
Rule-Utilitarianism is at least indeterminate with respect to, or at worst
excludes, minimizing-conditions. It can thus be regarded as self-defeating for a
wide range of cases in which acceptance of the principle is not general. For in
such cases, the point of (IRU), which is presumably also that of maximizing
utility, will not be achieved, since following (IRU) will not necessarily result in
the best consequences.
This suggests, incidentally, that the general coherence test is not a necessary
condition of a principle's being acceptable. For the results of the non-general
coherence argument may conflict with the results of the general coherence
argument. Alternative conclusions which can be drawn are, first, that both test
conditions are necessary (then perhaps no such principle is acceptable), or,
second, that the same point cannot be attributed to both principles.
We can avoid viewing the argument as substantive by viewing it as self-

referential. A principle can be assessed according to some criterion supplied by
reference to the principle itself. The general coherence argument then rests
upon the notion that an acceptable moral principle is coherent in the sense that
its general acceptance will not defeat its point.
What do we mean by ‘the point of a principle? Baier, for example, uses the
notion of a ‘purpose’: he says that a rule (or maxim) is self-frustrating and
therefore unacceptable (unfit to be universally taught) if its ‘purpose is
frustrated as soon as everybody acts on’ it (pp. 196–7). But what are we to count
as, and how do we determine, the purpose of a principle? Strictly speaking, a
principle (or rule or maxim) has no purpose. It is adopted for a purpose,
perhaps, used for some purpose, and so on. It seems odd, however, to attribute a
purpose (point, aim, goal, or end) to a principle (or rule or maxim) itself.
Page 33 of 38
RULE-UTILITARIANISM
This is not a verbal point. Consider two suggestions made by Harrison in his
version of the general coherence argument. He (p.155) speaks, apparently
indifferently, of ‘the purpose of the people who applied’ a given principle and of
‘the ends which determine them to adopt it’ (p. 132). But neither of these will
clearly do, nor are they obviously equivalents. The purpose of one who applies a
given principle, i.e. in applying it, need not be the same as the ends which
determine him to adopt it. Moreover, we can have many different purposes,
ends, aims, or goals in applying a given principle, in adopting it, and so on. Since
we are dealing with the acceptance of purportedly moral principles, however, we
might be properly concerned only with high-minded purposes, &c. If so, we can
restrict our attention to such as the following. One's purpose might be simply to
do right and not wrong, and to try to do this by acting on the basis of a principle
which one regarded as moral. (Are there any more appropriate purposes?) One's
reason for accepting it may be, not that it is a utilitarian principle as such, or
that its point is maximizing utility (one of which Harrison seems to suggest), but
that it seems the most viable alternative among the candidates.
But perhaps this does not help in the present case. Perhaps we should begin by
assuming that maximizing utility is the point of Act-Utilitarianism and proceed
from there. I think then that what must be meant by ‘the point of’ a principle is
the end, aim, purpose, or goal to which one is logically committed in accepting
the principle in virtue of its having certain features. When one accepts Act-
Utilitarianism as an ultimate, overriding moral principle—which is the only way,
it seems, one can fully accept it—one is logically committed to maximizing utility.
One is committed to applying the criterion of the best consequences in assessing
acts. One is committed to trying always to produce the best possible state of
affairs.
But in this sense, maximizing utility can be regarded as the point of (AU) as a
consequence of that principle's being a teleological (utilitarian) principle, and, in
particular, one of a certain species, e.g. positive (concerning the maximization of
utility), and not negative (concerning merely the minimization of suffering, &c).
And there seems no similar respect in which a non-teleological (non-utilitarian)
principle can be ascribed a point at all. This entails a further limitation upon the
general coherence argument. Such a self-referential test is not applicable to all
(p.156) kinds of candidate moral principles. Thus, passing the test cannot be
regarded as a necessary condition of a principle's being a moral principle or of
its being acceptable. Non-teleological principles cannot pass this test, but that is
not a mark against them. For they can neither pass nor fail it. As we shall see,
however, the fact that the test cannot be directly applied to non-teleological
principles does not imply that they are irrelevant to it.
Page 34 of 38
RULE-UTILITARIANISM
In what sense is the general coherence argument comparative? In the first

place, some standard is required for determining whether a given principle is
self-defeating. Consider Act-Utilitarianism: that principle is considered to be self-
defeating because utility would not be maximized by its general acceptance. But
what is it, not to maximize utility in this context? We know what it is in particular
cases: to effect a worse state of affairs than could be brought about if we had
acted otherwise. Hence the notion of maximizing utility has something to do with
the notion of an alternative.
It might be felt that a principle is self-defeating (in relation to maximizing utility)

merely if better states of affairs could possibly be brought about than those
which would be brought about by its general acceptance—in other words, if its
acceptance-utility is not a maximum. But this may be too strong a condition, for,
after all, it is possible that no principle has a maximum acceptance-utility, that
is, that no principle is such that, if it were generally accepted, then.the best
possible states of affairs would be effected. If we do not consider this too strong
a condition, we must acknowledge the possibility that every utilitarian principle
(or every principle having the point of maximizing utility) is ‘self-defeating’.
A related but somewhat different way of interpreting the test—another way of

viewing the comparison to be made—is to assess principles on the basis of
differences between their respective following-and acceptance-utilities. This
would make the test work against Act-Utilitarianism because the defect of (AU)
is supposed to be precisely that its acceptance-utility is lower than its following-
utility. In the case of (AU), of course, this is to say that its acceptance-utility is
not a maximum. In the case of principles that are not extensionally equivalent to
(AU), however, their following-utilities will not be as great as the following-utility
(p.157) of (AU). None the less, their acceptance-utilities may be even lower.
Indeed, this may be the case for Ideal Rule-Utilitarianism and other such
principles, as I shall argue. Thus, if we view the general coherence argument in
this way, it may again work against every utilitarian principle.
There is a further way of understanding the general coherence argument which I

would like to propose. Consider the alternative to general acceptance of Act-
Utilitarianism. It is the general acceptance of some other principle that can
presumably play the same role as a determinant of rightness and wrongness.
Thus, we could compare the extent to which general acceptance of a given
principle achieves its point with the extent to which general acceptance of other
principles achieves the same point. In the case of Act-Utilitarianism, assuming
the point to be that of maximizing utility, we would compare the acceptance-
utility of (AU) with the acceptance-utility of alternative principles.
Page 35 of 38
RULE-UTILITARIANISM
Now there are two ways of determining the range of comparisons to be made.
We could, on the one hand, compare the acceptance-utilities only of principles
having the same point-in this case, only of principles whose point is maximizing
utility. Among the principles to be compared with Act-Utilitarianism, therefore,
would presumably be Ideal Rule-Utilitarianism. But it should he observed that,
although many other principles do not have the same point as (AU), they can be
assigned acceptance-utilities. We could then broaden the range of comparison
and compare the acceptance-utility of (AU) with that of any alternative moral
principle.
If we pursue the argument along the latter lines, allowing a broad, essentially
unlimited class of comparisons, there develops the possibility that no principle
having the point of maximizing utility will be graded ‘acceptable’. For it is
possible that the acceptance-utility of some non-utilitarian principle is greater
than the acceptance-utility of every utilitarian principle. This would constitute a
paradox of utilitarianism. If it were possible to carry out such a test, it might be
worthwhile doing so, just to discover whether such a paradox can be confirmed.
But even if we were to compare only principles having the same point, it does
not appear that Ideal Rule-Utilitarianism would prove more acceptable than Act-
Utilitarianism. This may be (p.158) obscured by the formulation of Ideal Rule-
Utilitarianism. For the set of ideal rules is defined precisely as having the
highest acceptance-utility. It might seem, therefore, that Ideal Rule-
Utilitarianism itself has the highest acceptance-utility and that it necessarily
would outshine Act-Utilitarianism in such a comparison. This is not however the
case. For the acceptance-utility of the principle, (IRU), need not be as high as
the acceptance-utility of the set of ideal rules determined by the principle. We
know what (IRU) is—but we do not know the details of the set of ideal rules. The
considerations determining this set of rules are extremely complex—more so
than those determining the primitive rules of (PRU). Thus one could accept and
understand (IRU) without having a very accurate notion of the set of ideal rules
and therefore one could make serious errors—understandable errors, no doubt,
but errors none the less—in trying to apply (IRU). Indeed, it may be harder to
determine the set of ideal rules and thereby the implications of (IRU) in many
cases than to determine the implications of (AU) or (GU). Therefore, the
principle (IRU) itself may have a very low acceptance-utility, perhaps lower than
that of Act-Utilitarianism.
All this makes the position of the Ideal Rule-Utilitarian quite ironical. Since.
(IRU) promises to be extremely difficult to apply correctly, Ideal Rule-
Utilitarianism seems to be a most impractical theory, not the sort of theory
which overcomes the practical obstacles to successful application of Act-
Utilitarianism. And yet the argument for (IRU) begins with and is based upon
these practical obstacles.
Page 36 of 38
RULE-UTILITARIANISM
Two final comments upon the general coherence argument as it now appears. A
question could be raised whether Act-Utilitarianism and Ideal Rule-
Utilitarianism can be assumed to have the same point. For the notion of
maximizing utility plays a significantly different role within the two theories. In
the case of Act-Utilitarianism, the notion of maximizing utility (producing the
best consequences) is the criterion applied to acts in determining their rightness
or wrongness. But not so for Ideal Rule-Utilitarianism. This criterion is applied
instead to a set of rules, and, moreover, in a significantly different way. For while
following Act-Utilitarianism will produce the best consequences in every case,
following Ideal Rule-Utilitarianism will not necessarily (p.159) produce the
best consequences in every case. In accepting (AU) one is committed to
producing the best effects in each action; but in accepting (IRU) one is not so
committed. And, since (AU) and (IRU) are not extensionally equivalent, in
accepting (IRU) one is committed to regarding as right some acts that do not
have as good consequences as every alternative, or to regarding as wrong some
acts that do not have worse consequences than some alternative; but in
accepting (AU) one is not so committed.
Moreover, I have strong misgivings about the kind of comparison between

principles required by a general coherence argument. The approach implicit in
the argument is, of course, radically different from the mode of comparison I
have employed elsewhere in this study. What I have done up to now—what it
seemed quite natural to do—was compare the actual implications of analogous
principles, that is, compare what we can correctly infer from them about right
and wrong. On the general coherence argument, however, we are required to
consider what might be incorrectly inferred from the principles in question, for
the practical obstacles to successful acceptance of (AU) include difficulties in
determining what the principle actually implies owing to a shortage of
information, and so on. But note the kinds of factors we must consider in making
such a test and therefore in assessing the ‘acceptability’ of a candidate moral
principle. We are to consider, in effect, the mistakes we make, the errors to
which we are prone, our weaknesses, temptations, passions and cravings, our
blockheadedness, ignorance, confusion and stupi-dity—all of which bear upon
what we might actually infer and do, as opposed to what we ought to infer and
do according to the principles in question. Theories are not normally compared
in such a way. No other kind of theory would be assessed on the grounds implicit
in the general coherence argument.
Page 37 of 38
RULE-UTILITARIANISM
The rule-utilitarian is free, of course, to view the general coherence argument in

various ways, as I have suggested. It appears, however, that none of these modes
of argument will count simultaneously against Act-Utilitarianism and for a
theory like Ideal Rule-Utilitarianism. Alternatively, the rule-utilitarian can take a
radically different line from the one I have outlined here. He can disavow any
intention of developing an alternative to (AU) that is utilitarian in the sense in
which (AU) is. He (p.160) can say that his theory is simply designed to
accommodate the counter-examples to Act-Utilitarianism and that it need not fit
any preconceived notions of a utilitarian theory. But in the first place, (IRU) does
not seem to satisfy the counter-examples against (AU)—as I shall re-emphasize
in the next chapter. And secondly, to take such a stand is to say, not that Ideal
Rule-Utilitarianism or some alternative is a better theory of the same kind, but
rather that it is a theory of another kind, and that the other is a better kind. But
this remains to be seen, especially since the differences between (IRU) and (AU)
still do not account for arguments from justice and fairness.
Access brought to you by:
Page 38 of 38

David Lyon - Rule Utilitarianism

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

David Lyon - Rule Utilitarianism

Загружено:

Авторское право:

Доступные форматы

RULE-UTILITARIANISM

University Press Scholarship Online

Oxford Scholarship Online

Forms and Limits of Utilitarianism

Print publication date: 1965

Abstract and Keywords

Keywords: utility, primitive rule-utilitarianism, moral rules, utilitarian generalization, Act-

Connected with this topic is the relation of utilitarian generalization to rule-

In order to discuss the various types of rule-utilitarianism, it is necessary to

In the case of Act-Utilitarianism, by contrast, rules can theoretically be

The fact that a type of rule-utilitarianism can be developed from utilitarian

Some qualifying clause is essential whenever such a judgement-rule is justified

What kind of qualification? Once we speak of presumptions which can be

This type of rule suggests a reformulation of utilitarian generalization into

I want now to indicate the alternative schema for theoretically complete

What I have in mind is this. Instead of mentioning the exception-making criteria

All the concise rules of a given primitive rule-utilitarianism (p.128) can in

I am especially interested in two classes of exempting-conditions that would be

I shall now try to indicate by means of an example the nature of maximizing-and

Let us now consider a minimizing-condition, using the same example. Suppose

Maximizing-and minimizing-conditions should be kept distinct. Although both

I should mention finally that what I am calling ‘minimizing-conditions’ seem to

Now some general comments on primitive rule-utilitarianism. In the first place,

It may be misleading to speak of a set of rules extensionally equivalent to one

Primitive rule-utilitarianisms must also be seen in the light of earlier

As I have already suggested, however, rule-utilitarians do have an alternative—

(p.134) These should not be viewed as modifications of primitive rule-

(p.135) Moreover, it seems reasonable to suppose that, in many (perhaps all)

(p.136) B. Non-Primitive Rule-Utilitarianism

The idiom commonly used in the formulation of rule-utilitarian theories is

(p.137) Similarly, its general utilitarian analogue becomes:

Brandt calls this a ‘specious’ rule-utilitarianism (‘Toward a Credible Form of

The difference is briefly this. Whereas minimizing-conditions can and must be

The point can be put in another way. It is misleading to analyse (PRU) in

How then are we correctly to formulate (PRU)? Perhaps this way:

It is understood, of course, that the case in question is always to be treated as

(p.140) A comparison of (PRU), and thereby of (AU), with Ideal Rule-

Ideal Rule-Utilitarianism (or a paradigm principle of this type) is formulated,

Rule-Utilitarians have been aware of this problem and sensitive to criticisms

The resulting principle is:

It should be observed, however, that the de facto morality of a given community

These observations suggest that the rule-utilitarian is gradually forced to forsake

C. Utility and Rules

(p.144) The question is: how is it possible for a utilitarian consistently to

A strictly analogous comparison could be made between non-primitive rule-

I shall now try to reconstruct one particularly relevant kind of argument by

Consider the argument in practical terms first. It is admitted that mere

But interestingly enough, the Act-Utilitarian is apt to accept this argument, to

In order to understand the Act-Utilitarian's position, however, in order to see

If we fail to keep this distinction in mind confusions arise. For example,

A second source of confusion is the attempt by some theorists (p.148) to

In response to those practical considerations showing that there is a risk in

But it must be emphasized that such a practical addendum to the Act-

The Act-Utilitarian's aim in recommending the adoption and observance of de

Primitive Rule-Utilitarianism is, as it were, half-way between Act-Utilitarianism

This is not, however, an appropriate rendering of the general coherence

The general coherence argument becomes more credible when we displace

(p.153) It is not uncommon for philosophers to propose necessary, presumably

I want to emphasize that the general coherence argument is formal because it is

If we understand the general coherence argument in this way, however, we seem

(p.154) Moreover, even within the limited context of utilitarianism, the

We can avoid viewing the argument as substantive by viewing it as self-

In what sense is the general coherence argument comparative? In the first